This is a report that assesses the ELR (Epidemiological Level of Risk) process, as performed by the Analytics and Intel teams of the COVID-19 epi response pillar.

1 Preparation

1.1 Packages

1.2 Import latest data

Manual selection of data, so as to not store these data in the public-facing repository.

1.3 Data cleaning

This analysis utilizes the last 6 weeks of ELR data, from 2021-06-07`

The latest date in the dataset is 2021-07-19. There are 1418 records.

2 Map of the most recent ELR classifications

4 Trajectories over time

The final ELR classification by country

## [1] "No Data"   "Minimal"   "Low"       "Medium"    "High"      "Very high"
## [7] "Critical"
## [1] TRUE
## Warning in to_lodes_form(data = data, axes = axis_ind, discern =
## params$discern): Some strata appear at multiple axes.

## Warning in to_lodes_form(data = data, axes = axis_ind, discern =
## params$discern): Some strata appear at multiple axes.

## Warning in to_lodes_form(data = data, axes = axis_ind, discern =
## params$discern): Some strata appear at multiple axes.

6 Early detection

We examine the proportion of countries in a given week that were detected by the ELR process prior to their becoming a hotspot. That is, for the week when a country was first classified as either 1 - Critical, 2 - Very high, were they listed in the previous two weeks as either: 1 - Critical, 2 - Very high, 3 - High, 4 - Medium ?

6.1 Trajectories of “Missed” countries

The “missed” countries were: Honduras, Iran (Islamic Republic of), Lao People’s Democratic Republic, Libya, Rwanda, Senegal

Below, we explore their ELR trajectories:

When we compare the ELR trajectories to the epidemic curves, it becomes clear that some of these countries were not truly “missed” but rather ignored for some reason without any ELR classification prior to their case surge.

## [[1]]

## 
## [[2]]

## 
## [[3]]

## 
## [[4]]

## 
## [[5]]

## 
## [[6]]

7 Characteristics of ELR “Medium”

This section examines the nature and trajectories of countries that are classified as ELR Medium.

If we plot the epidemiological metrics of countries when they are first classified as “Medium”, they tend to be either:

  • High incidence and low growth (chronic - e.g. many countries in South America), or
  • Low incidence and high growth (budding outbreaks)

The plot below shows this dynamic, as most Medium countries are either in the upper-left, or the lower-right.

The next question is - what happens to these countries after they are first classified as “Medium”? Do they rise upward to Critical or Very High ELR classifications? Or do they stay lower, at Medium or below?

The plot below depicts the ELR classification trajectories of countries before and after they are first classified as “Medium” (day 0 on the x-axis). Each tiny line is one country’s ELR trajectory, and there is a smoothed large line. There are two distinct patterns:

  1. Countries that increase to Critical or Very High within a few weeks (red)
  2. Countries that remain at lower ELR levels (blue)

Circling back to that original points plot of the metrics of metrics when they are first classified as Medium…

While some countries in both the upper-left and lower-right sectors did increase to Critical/Very High ELR within a few weeks of being classified as Medium (the red points above), countries in the upper-left sector (with higher incidence and lower growth) seemed to be more likely to increase than countries in the lower-right sector.

So we may want to pay special attention to Medium countries that have higher incidence, even if they have lower growth rates.

One hypothesis could be that countries with smaller but faster-growing epidemics are increasing testing and so their growth rates are high. But their overall incidence is still relatively low.

8 Global overview